Log in

AI & Software

Where AI spending is actually landing

Model releases, enterprise adoption and software margins, read for what they do to revenue — across the labs, the platforms and the software they run on.

What Big Funds Own42 fundsWhat Berkshire, Bridgewater and other big funds reported holding last quarter.What Analysts Expect16 stocksRatings and price targets from the big institutions, each with its public source.

2026-08-10

Meta's Glimmer reframes AI capex as “agent infrastructure,” not a bigger chatbot insight cover
Industry News
META•8 min read

Meta's Glimmer reframes AI capex as “agent infrastructure,” not a bigger chatbot

Meta is tying ~130B–145B capex guidance in 2026 to “AI efforts,” while launching Muse Glimmer as an open-weight, ~30B-parameter model optimized for always-on local agent workflows. The strategic implication is that Meta’s spend is increasingly rationalized by an always-running, user-specific agent loop—changing how compute, data, and distribution translate into revenue versus pure chatbot demos.

TTM revenue: $228.2BTTM R&D expense: $71.6B
Read article
Microsoft's September Maia 300 reveal tests whether custom inference silicon can undercut NVIDIA without breaking the supply chain insight cover
Industry News
MSFT · NVDA•8 min read

Microsoft's September Maia 300 reveal tests whether custom inference silicon can undercut NVIDIA without breaking the supply chain

Microsoft is signaling a next-generation custom AI accelerator (“Maia 300”) for a September public unveiling and 2027 deliveries. Its earlier Maia 200 already claims 30% better inference performance-per-dollar, so the real question isn’t specs—it’s whether Microsoft can scale volume enough to shift the inference cost curve and take share from NVIDIA’s full-stack economics.

Maia 200: performance-per-dollar claim: 30% betterMaia 200: scale-up cluster support: 6,144 accelerators
Read article

2026-08-09

The “AI Safety Test” Is Turning into a Security & Disclosure Liability insight cover
Private Company
MSFT · NVDA•9 min read

The “AI Safety Test” Is Turning into a Security & Disclosure Liability

When governments and frontier labs rely on opaque evaluation/test pathways, the act of testing can create new cyber and reporting attack surfaces. The compliance stack is being forced to move from “did we test?” to “how do we safely disclose, coordinate, and audit the tests themselves?”

Microsoft's FY2025 revenue (proxy for ca: $281.7BMicrosoft's FY2025 net income (ability t: $101.8B
Read article
Airbnb's 60% AI-Coded Engine Turns the Travel Stack Into an Agent-First Listings War insight cover
Industry News
ABNB · BKNG•20 min read

Airbnb's 60% AI-Coded Engine Turns the Travel Stack Into an Agent-First Listings War

Airbnb is testing a natural-language AI search toggle while CEO Brian Chesky discloses that nearly 60% of new code is now AI-written — a productivity gap that lets the platform ship AI features faster than Booking Holdings or Expedia Group can copy them. The same shift forces every host to optimize listings for AI agents, not just guests, rewriting host SEO economics. With $27.2B in Q2 gross booking value at a 13.2% take rate and 35% adjusted EBITDA margins, Airbnb's AI velocity is the moat the incumbents have to chase, not the other way around.

Airbnb Q2 2026 GBV: $27.2BAirbnb Q2 2026 take rate: 13.2%
Read article
Anthropic’s “global affairs” hire is the clearest proof frontier AI is being regulated like a geopolitical asset insight cover
Private Company
7 min read

Anthropic’s “global affairs” hire is the clearest proof frontier AI is being regulated like a geopolitical asset

Anthropic’s first Chief Global Affairs Officer, former California Supreme Court justice Mariano-Florentino “Tino” Cuéllar, formalizes something markets usually underwrite implicitly: government access and compliance bargaining. The appointment matters because it reframes frontier AI risk as a relationship-and-contract problem—after Anthropic has already faced export-control actions and helped shape California’s frontier-AI rules like SB 53.

Export-control outcome: Access suspension for foreign naGovernance framework: SB 53 endorsed by Anthropic
Read article
Apple's CXMT test isn’t a price story—it’s a sanctions-and-qualification story that can *stabilize Apple’s bill or strengthen SK hynix and Samsung’s pricing power insight cover
Industry News
AAPL · SKHY•9 min read

Apple's CXMT test isn’t a price story—it’s a sanctions-and-qualification story that can *stabilize Apple’s bill or strengthen SK hynix and Samsung’s pricing power

Apple is reported to have progressed to qualification testing of [CXMT] PRC DRAM for iPhone/Mac devices sold in China, while also engaging US policy for clearance. The first-order market implication is that CXMT can’t credibly undercut pricing yet, so the substitution “math” likely sharpens near-term leverage for SK Hynix and Samsung Electronics while raising a binary compliance risk for Apple’s sourcing pipeline.

Apple revenue (TTM): $466.8BApple EBIT margin (TTM): 33.2%
Read article
Cloudflare’s Q2 AI-Inference Guide Raised the “Toll” Value of the Internet Edge—So NET Repriced From CDN to Answer-Engine Infrastructure insight cover
Earnings
NET•7 min read

Cloudflare’s Q2 AI-Inference Guide Raised the “Toll” Value of the Internet Edge—So NET Repriced From CDN to Answer-Engine Infrastructure

Cloudflare used its Q2 2026 print to guide fiscal 2026 revenue to $2.864B–$2.870B and non-GAAP EPS to $1.25–$1.26, a combination the market treated like proof that AI inference traffic is accelerating at the edge. The read-through is that Cloudflare (and edge rivals) increasingly function as the “metering + policy control plane” for machine-to-machine traffic—not just a caching layer.

Q2 2026 revenue: $696.1MFY 2026 revenue guide: $2.864B–$2.870B
Read article
KOSPI volatility ebbed—but the AI-memory trade is now a different risk asset after leveraged flush insight cover
Markets / Event
7 min read

KOSPI volatility ebbed—but the AI-memory trade is now a different risk asset after leveraged flush

South Korea’s volatility gauge has fallen to a two-month low after forced liquidations and regulatory tightening on single-stock leveraged ETFs. The key implication for investors isn’t “risk-on returns”—it’s that the AI-memory complex’s tape is now shaped more by de-leveraging mechanics than by momentum alone.

Volatility index (June peak): 96.9Volatility index (end state): Two-month low
Read article

2026-08-08

Apple turns Alibaba Qwen into a China Mac distribution channel—iPhone was the pilot insight cover
Industry News
AAPL · BABA•8 min read

Apple turns Alibaba Qwen into a China Mac distribution channel—iPhone was the pilot

Reuters says Apple’s China rollout lets eligible Mac users connect to Alibaba’s Qwen service via a macOS extension (macOS 26.6+), routing Qwen through Siri and Writing Tools for more detailed responses. That shifts the “Qwen for Apple Intelligence in China” story from a mobile-only monetization channel to desktop/enterprise workflows—raising the odds that Alibaba captures more downstream AI usage time (and thus higher Cloud/enterprise pull) alongside Apple’s China Services stickiness.

Mac shipments (mainland China, Q1): ~800,000 unitsYoY change (Mac shipments, Q1): -9%
Read article
AppLovin's Q2 miss broke the “open access = upside” story—because its own model-driven auction timing still controls monetization insight cover
Earnings
APP•7 min read

AppLovin's Q2 miss broke the “open access = upside” story—because its own model-driven auction timing still controls monetization

AppLovin didn’t miss on demand; it missed on timing, with management attributing the softer quarter to lighter-than-normal model improvement. That matters to the “open access / distribution” thesis because ad-tech economics clear through pricing/take-rate via auctions, not just incremental reach—so one weak quarter can vaporize the narrative even when the machine keeps growing.

Q2 2026 Revenue: $1.924BQ1 2026 Revenue: $1.842B
Read article
Cloudflare turns the browser into agent compute—then prices the cost drivers (CPU/memory) inside Workers P&L insight cover
Industry News
NET•9 min read

Cloudflare turns the browser into agent compute—then prices the cost drivers (CPU/memory) inside Workers P&L

With Cloudflare’s Kitesurf (launched Aug 6), the “browser” becomes a stateless Rust engine running inside Workers, optimized for AI agent tasks rather than human browsing. The key economic tell is that Cloudflare reports materially lower CPU and especially memory than Chromium for common agent actions—shifting Cloudflare’s agent-compute revenue mechanics away from CDN/security toward browser-like workloads metered by resource cost.

Revenue (FY2023 → FY2025): $1.30B → $2.17BNet income (FY2023 → FY2025): -$184M → -$102M
Read article
Nvidia’s $3B Lancium Bet Turns Stargate Into a Power-Site Landlord Play—Not a GPU Story insight cover
Industry News
8 min read

Nvidia’s $3B Lancium Bet Turns Stargate Into a Power-Site Landlord Play—Not a GPU Story

The key market constraint behind Stargate-style AI buildouts is shifting from GPU availability to multi-year power + grid interconnect lock-up. Nvidia’s $3B-scale equity commitment to Lancium (the power infrastructure developer tied to Stargate) matters because it anchors the “AI grid” that determines where compute can actually be deployed.

Stargate scale (planned capacity + investment wi: Nearly 7GW; $400B+Stargate end-goal: $500B; 10GW
Read article
OpenAI’s “Critical Cyber” threshold turns AI-safety into a security-vendor compliance stack—creating a new 30-day-style review process even without a “kill switch” law insight cover
Private Company
11 min read

OpenAI’s “Critical Cyber” threshold turns AI-safety into a security-vendor compliance stack—creating a new 30-day-style review process even without a “kill switch” law

OpenAI’s Aug 7 post on “Responding to the next frontier of critical cyber capabilities” defines when a frontier model can perform end-to-end novel cyberattack strategies, and says it is strengthening/pausing releases and agent workloads accordingly. That threshold reframes AI safety from “model behavior promises” into a test-and-controls gate—one that can force security vendors and evaluators to operationalize compliance via measured exploitability and environment isolation, not just policy statements.

Palo Alto Networks (PANW) latest quarter revenue: $10.61BCrowdStrike (CRWD) latest quarter revenue (TTM d: $5.09B
Read article
[OpenAI]’s [NextSlide] acquisition signals the productivity-stack era: ChatGPT is becoming the “AI-native Office distribution layer” insight cover
Private Company
8 min read

[OpenAI]’s [NextSlide] acquisition signals the productivity-stack era: ChatGPT is becoming the “AI-native Office distribution layer”

OpenAI’s Aug 2026 move to bring the [NextSlide] presentation team into the [OpenAI]/[ChatGPT] product suggests the next growth frontier is vertical workflow SaaS—starting with the work of turning ideas into decks. The deal’s disclosed specifics are thin, but the direction is testable: Microsoft and Alphabet can’t defend Office and Workspace UX with model access alone, they must out-execute on AI-native creation and formatting loops.

Microsoft revenue (TTM snapshot): $331.8BAlphabet revenue (TTM snapshot): $445.9B
Read article
OpenAI’s $300–$400 Smart Speaker Trade: a voice-first margin switch that forces retailers and phone ecosystems to pay for access insight cover
Private Company
AAPL · SONO•9 min read

OpenAI’s $300–$400 Smart Speaker Trade: a voice-first margin switch that forces retailers and phone ecosystems to pay for access

A leaked report places OpenAI’s first consumer device—a screenless, donut-shaped AI speaker—at ~$300–$400 and suggests a 2027 entry. For investors, the key question isn’t “will people buy it?” but “who funds the margin bridge from OpenAI API to an always-listening voice edge in the home,” with likely knock-on pressure for Apple (ecosystem control) and Sonos (connected-audio relevance) while Amazon monetizes shipping distribution.

Reported unit price range: $300–$400Reported first launch window: 2027
Read article
Rippling turns enterprise AI ROI into an auditable cost center, not a productivity vibe insight cover
Private Company
RIPPLING•8 min read

Rippling turns enterprise AI ROI into an auditable cost center, not a productivity vibe

Rippling says it burned “millions” on AI tokens in a few months, then built an internal employee-ROI tracker—now launched as AI Spend Console—to connect token spend to measurable outcomes. The move reframes AI from a blanket capability upgrade into a controllable spend-and-output system, forcing HR/payroll and adjacent enterprise software vendors to prove AI value in employee hours, not marketing.

Token spend growth: 80% MoMEmployee concentration: 10–15%
Read article
The “AI Capex Reality Check”: a $400M bet on a stealth chip-fab startup after a $16B unwind is a signal about who funds the 2027–2030 buildout insight cover
Private Company
9 min read

The “AI Capex Reality Check”: a $400M bet on a stealth chip-fab startup after a $16B unwind is a signal about who funds the 2027–2030 buildout

A post-bailout $400M commitment by Leopold Aschenbrenner’s Situational Awareness into Source Foundry implies that even after a leverage-driven AI-equity unwind, the fund still believes the frontier-AI capex cycle will require new manufacturing capability—especially in tools and process. Public-market drawdowns may have repriced the equity trade, but the chip-supply-chain cash allocation appears to be moving toward the “picks-and-shovels” of scaling production rather than competing on model IP.

ASML (tooling proxy): Revenue: $35.3B (TTM)ASML (profit engine): Operating margin: 35.4% (TTM)
Read article

2026-08-07

Alibaba’s “pay-for-heavy-users” open-weight twist turns Qwen into a monetization moat—weeks after the open-weights cost wave started insight cover
Industry News
8 min read

Alibaba’s “pay-for-heavy-users” open-weight twist turns Qwen into a monetization moat—weeks after the open-weights cost wave started

Reuters reports Alibaba plans to charge “big users” for its next open-weight model (Qwen3.8-Max), with timing aimed for next week and revenue-sharing for commercial “sale-as-a-service” deployments. The immediate investor signal: if open-weight can be turned into a paid tail without killing developer adoption, it pressures US hyperscalers’ inference-at-low-margin playbooks that currently benefit from free distribution.

Reported model targeted: Qwen3.8-MaxReported commercialization gate: Sale-as-a-service
Read article
Advanced Micro Devices' first “one-model-to-silicon” buy re-ranks the inference chip war insight cover
Industry News
AMD•7 min read

Advanced Micro Devices' first “one-model-to-silicon” buy re-ranks the inference chip war

By acquiring Taalas, Advanced Micro Devices is buying an inference architecture where the model is hardwired into custom silicon, not just accelerated at runtime. That shifts the hyperscaler vendor map toward “model-specific throughput” players, while pressuring GPU-style margins to defend against a new cost-performance axis.

FY revenue: $34.6BFY net income: $4.27B
Read article
Anthropic’s 85% fewer biology fallbacks reframes biosecurity as a measurable, cost-to-serve moat insight cover
Private Company
7 min read

Anthropic’s 85% fewer biology fallbacks reframes biosecurity as a measurable, cost-to-serve moat

Anthropic says its Aug 7, 2026 update cut biology-related “fallbacks” by about 85% in testing by tightening the boundary of its biology safeguards (constitutional classifier tuning). If regulated labs buy based on operational friction—not just policy—this kind of measurable false-positive reduction can lower reroute-and-retry costs while improving throughput into life-science workflows.

Biology-related fallbacks (testing): ~85%Total fallbacks (directional product mix effect): Down materially
Read article

What to expect

Evidence-first notes with a visible point of view.

This section collects sharp takes on earnings, shareholder meetings, and market structure. Each new piece should make the thesis, the facts, and the implications obvious within the first few screens.

Expect direct analysis, not generic commentary.

Expect the data to be explicit and the argument to be easy to follow.